Received-SPF: pass (google.com: domain of linux-kernel+bounces-205823-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Date: Fri, 7 Jun 2024 12:30:43 +0200
From: Peter Zijlstra <peterz@infradead.org>
To: Zhang Qiao <zhangqiao22@huawei.com>
Cc: mingo@redhat.com, juri.lelli@redhat.com, vincent.guittot@linaro.org,
	dietmar.eggemann@arm.com, rostedt@goodmis.org, bsegall@google.com,
	mgorman@suse.de, bristot@redhat.com, vschneid@redhat.com,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] sched: Initialize the vruntime of a new task when it is
 first enqueued
Message-ID: <20240607103043.GO8774@noisy.programming.kicks-ass.net>
References: <20240606121133.2218723-1-zhangqiao22@huawei.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20240606121133.2218723-1-zhangqiao22@huawei.com>

On Thu, Jun 06, 2024 at 08:11:33PM +0800, Zhang Qiao wrote:
> When create a new task, we initialize vruntime of the new task
> at sched_cgroup_fork(). However, the timing of executing this
> action is too early and may not be accurate.
> 
> Because it use current cpu to init the vruntime, but the new
> task actually runs on the cpu which be assigned at wake_up_new_task().
> 
> To optimize this case, we pass ENQUEUE_INITIAL flag to
> activate_task() in wake_up_new_task(), in this way,
> when place_entity is called in enqueue_entity(), the
> vruntime of the new task will be initialized. At the same
> time, place_entity in task_fork_fair() is useless, remove it.

The better argument would've looked at history to see why the code was
the way it is and then verify those reasons are no longer valid.

Specifically, I think these are remains of child_runs_first, and that is
now gone.

Can you verify and update accordingly?

> Signed-off-by: Zhang Qiao <zhangqiao22@huawei.com>
> ---
>  kernel/sched/core.c |  2 +-
>  kernel/sched/fair.c | 16 ----------------
>  2 files changed, 1 insertion(+), 17 deletions(-)
> 
> diff --git a/kernel/sched/core.c b/kernel/sched/core.c
> index bcf2c4cc0522..b4ff595a2dc8 100644
> --- a/kernel/sched/core.c
> +++ b/kernel/sched/core.c
> @@ -4897,7 +4897,7 @@ void wake_up_new_task(struct task_struct *p)
>  	update_rq_clock(rq);
>  	post_init_entity_util_avg(p);
>  
> -	activate_task(rq, p, ENQUEUE_NOCLOCK);
> +	activate_task(rq, p, ENQUEUE_NOCLOCK | ENQUEUE_INITIAL);
>  	trace_sched_wakeup_new(p);
>  	wakeup_preempt(rq, p, WF_FORK);
>  #ifdef CONFIG_SMP
> diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
> index efce2d36a783..bb5f376fd51e 100644
> --- a/kernel/sched/fair.c
> +++ b/kernel/sched/fair.c
> @@ -12702,23 +12702,7 @@ static void task_tick_fair(struct rq *rq, struct task_struct *curr, int queued)
>   */
>  static void task_fork_fair(struct task_struct *p)
>  {
> -	struct sched_entity *se = &p->se, *curr;
> -	struct cfs_rq *cfs_rq;
> -	struct rq *rq = this_rq();
> -	struct rq_flags rf;
> -
> -	rq_lock(rq, &rf);
> -
>  	set_task_max_allowed_capacity(p);
> -
> -	cfs_rq = task_cfs_rq(current);
> -	curr = cfs_rq->curr;
> -	if (curr) {
> -		update_rq_clock(rq);
> -		update_curr(cfs_rq);
> -	}
> -	place_entity(cfs_rq, se, ENQUEUE_INITIAL);
> -	rq_unlock(rq, &rf);
>  }
>  
>  /*
> -- 
> 2.18.0.huawei.25
>